The estimated risk for language impairment (LI) in monolingual English speaking children is about 7%. A model for identifying language impairment, referred to as the EpiSLI system (Tomblin, Records, & Zhang, 1996), yields relatively high levels of specificity and sensitivity for identification of LI in kindergarten children from monolingual English speaking backgrounds. Systematic study is needed to determine a) the extent to which level of proficiency in English is related to risk for misidentification of language impairment, b) if the EpiSLI criteria can be applied to a new bilingual language test, c) to what extent testing (using a variety of assessment tasks) in English and the native language can be used to identify language impairment, and d) potential differences in the way bilingual learner's (language impaired vs. typical) performance on language measures changes with increased exposure to English. Due to the shortage of bilingual clinicians and the diversity of languages spoken in the US it would be helpful to determine the circumstances in which standardized testing, language learning measures, and/or processing measures administered in English can be used to reliably identify language impairment in bilingual children. We have planned a three phase study to address these issues. In phase 1,1500 Latino children will participate in language screening to determine the levels of risk across the bilingual population. In phase 2, a comparison of performance on standardized measures in Spanish and English will be administered to the at risk bilingual children identified in phase 1 to determine if the Epi SLI cut off values hold for bilinguals. Further, processing and learning measures will be administered in English, to explore the extent to which such measures accurately predict language impairment at various levels of English proficiency. In phase 3, children will participate in follow up testing evaluating language growth. Assessment of the persistence of language impairment across a 1-year period will help determine whether increase in English language skills affects classification accuracy of language impairment when measures are utilized in English. [unreadable] [unreadable]